How to get rocm running on R9 390 series?

[40476]

Your question

Been pulling my hair out (figuratively of course) on how to get ROCm running on R9 390, I know its like 10 years old but my only problem with it is that I cannot get ROCm to work on it, I am open to alternative solutions to getting comfyUI on my GPU.

Logs

usr_40476@2xeon:~/APPS/ComfyUI> python3.12 main.py 
Traceback (most recent call last):
  File "/home/usr_40476/APPS/ComfyUI/main.py", line 91, in <module>
    import execution
  File "/home/usr_40476/APPS/ComfyUI/execution.py", line 13, in <module>
    import nodes
  File "/home/usr_40476/APPS/ComfyUI/nodes.py", line 21, in <module>
    import comfy.diffusers_load
  File "/home/usr_40476/APPS/ComfyUI/comfy/diffusers_load.py", line 3, in <module>
    import comfy.sd
  File "/home/usr_40476/APPS/ComfyUI/comfy/sd.py", line 5, in <module>
    from comfy import model_management
  File "/home/usr_40476/APPS/ComfyUI/comfy/model_management.py", line 143, in <module>
    total_vram = get_total_memory(get_torch_device()) / (1024 * 1024)
                                  ^^^^^^^^^^^^^^^^^^
  File "/home/usr_40476/APPS/ComfyUI/comfy/model_management.py", line 112, in get_torch_device
    return torch.device(torch.cuda.current_device())
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/usr_40476/.local/share/pipx/venvs/pip/lib64/python3.12/site-packages/torch/cuda/__init__.py", line 971, in current_device
    _lazy_init()
  File "/home/usr_40476/.local/share/pipx/venvs/pip/lib64/python3.12/site-packages/torch/cuda/__init__.py", line 319, in _lazy_init
    torch._C._cuda_init()
RuntimeError: No HIP GPUs are available
usr_40476@2xeon:~/APPS/ComfyUI>

Other

No response

[patientx]

you need rocm libraries for a start but as far as I know there are none for your gpu. Your gpu's code is : GFX7 (gfx702) . So far we have (for various rocm versions from 5.7.1 to 6.2.x) 803,900,902,906,1010,1011,1012,1030,1031,1032,1035,1100,1101,1102,1103,1150. The rest are either too old , yours is from 7th as you can see , nothing from there. Or too new that they don't need libraries to run rocm. The only option you have is directml.

[40476]

working on installing directml right now

[40476]

turns out directml is not supported on linux unless wsl, so does that mean i am just out of luck?

[patientx]

turns out directml is not supported on linux unless wsl, so does that mean i am just out of luck?

windows :(